bioRxiv Science⌕ Search

Biology subjects

Mei, W.

Publications and source records attributed to Mei, W..

13 recordsLinked to original sources
bioRxiv2026

Evaluating agentic AI for biological discovery in autonomous and copilot settings

Advances in large language models (LLMs)-based artificial intelligence (AI) agents have improved their ability to execute structured analytical workflows, including standard bioinformatic pipelines for biological discovery. However, computational biology rarely consists of deterministic pipeline execution alone. Biological datasets are heterogeneous and noisy, and meaningful discovery often requires open-ended hypothesis generation and iterative reasoning over multimodal evidence. These challenges are particularly evident in multi-omic studies, where paired molecular modalities and heterogeneous clinical contexts create both opportunities and obstacles for discovery. The extent to which emerging agentic AI systems can support or automate this mode of scientific discovery remains poorly understood. Here, we systematically evaluated the capabilities and limitations of agentic AI for biological discovery using multi-omic single cell datasets spanning 11 cancer types. We developed the Multistep Multimodal Multiomic Agentic (M3A) Framework to support LLM-driven reasoning over persistent multimodal data states and to capture agentic reasoning behavior in autonomous and human-AI copilot settings. Using this framework, we assessed AI agents across complementary tasks, including autonomous cell-type annotation, generation of falsifiable biological hypotheses from gene programs, and copilot experiments testing the effect of human involvement and domain expertise. We found that current AI agents are effective at broad, systemic exploration of complex data, whereas domain experts remain critical for methodological guidance and biological synthesis across analyses. Together, our results delineate the current potential and boundaries of agentic AI in computational biology, and establish a framework for evaluating AI systems designed to support biological discovery.

cancer biology↗
bioRxiv2026

Bacteroides intestinalis-Driven Arabinoxylan Fermentation Mitigates Inflammatory and Metabolic Dysfunction

Insufficient dietary fiber intake is closely linked to gut microbiome dysfunction and increased risk of noncommunicable diseases. Synergistic synbiotics, pairing defined microbes with their dietary substrates, offer a precise strategy to restore microbiome function. Here, we show that pairing the human colonic commensal Bacteroides intestinalis with insoluble wheat arabinoxylan (inWAX) yields pronounced metabolic and anti-inflammatory benefits. Using mono-associated gnotobiotic mice and high-fat diet-induced obese mice, we demonstrate that this synbiotic enhances resistance to intestinal inflammation and improves glucose homeostasis. Mechanistically, this synbiotic increases production of 6-hydroxylated bile acids with anti-diabetic and anti-steatotic effects. It also promotes microbial transformation of phenolic compounds and redirects protein metabolism toward neuroactive metabolites. These metabolic shifts are accompanied by transcriptional remodeling in the colon, spleen and liver, involving induction of genes essential for circadian rhythm, lipid metabolism, immune defense, and bile acid 6-hydroxylation. Notably, Y chromosome-linked genes associated with epigenetic regulation and protein turnover are also induced, suggesting a potential sex-specific response to synbiotics. Together, our findings establish a mechanistic framework for targeted synbiotic interventions to combat inflammatory and metabolic disorders.

microbiology↗
bioRxiv2025

Inhibiting RhoA Activation via GDP-state Stabilization to Relieve Heart Failure

High rates of heart failure (HF) morbidity and mortality have made targeting myocardial remodeling--particularly hypertrophy and fibrosis--a key therapeutic focus. RhoA, which regulates cytoskeletal reorganization and cell migration, plays a role in this process. However, RhoA has long been considered "undruggable", due to its strong binding to its endogenous substrates, GDP/GTP, and the lack of well-defined pockets for drug targeting. Here, we discovered a cryptic pocket proximate to GDP within RhoA and identified a natural product, AH001, binds here and interacts with GDP, stabilizing RhoAs interaction with its endogenous inhibitor, RhoGDI. AH001 reduced the downstream MRTFA nuclear translocation and downregulated fibrosis/hypertrophy proteins. Consequently, AH001 mitigated myocardial remodeling in multiple HF animal models, and in the 3D myocardial tissue model. Our findings highlight the therapeutic potential of inhibiting RhoA activation in myocardial remodeling, ultimately targeting HF, and offer a promising avenue for developing reversible inhibitors against undruggable GTPases.

pharmacology and toxicology↗
bioRxiv2024

PTBP1 mediates Sertoli cell actin cytoskeleton organization by regulating alternative splicing of actin regulators

Spermatogenesis is a biological process within the testis that produces haploid spermatozoa for the continuity of species. Sertoli cells are somatic cells in the seminiferous epithelium that orchestrate spermatogenesis. Cyclic reorganization of Sertoli cell actin cytoskeleton is vital for spermatogenesis, but the underlying mechanism remains largely unclear. Here, we report that RNA-binding protein PTBP1 controls Sertoli cell actin cytoskeleton reorganization by programming alternative splicing of actin cytoskeleton regulators. This splicing control enables ectoplasmic specializations, the actin-based adhesion junctions, to maintain the blood-testis barrier and support spermatid transport and transformation. Particularly, we show that PTBP1 promotes actin bundle formation by repressing the inclusion of exon 14 of Tnik, a kinase present at the ectoplasmic specialization. Our results thus reveal a novel mechanism wherein Sertoli cell actin cytoskeleton dynamics is controlled post-transcriptionally by utilizing functionally distinct isoforms of actin regulatory proteins, and PTBP1 is a critical regulatory factor in generating such isoforms.

molecular biology↗
bioRxiv2024

Dzip1 is dynamically expressed in the vertebrate germline and regulates the development of Xenopus primordial germ cells.

Primordial germ cells (PGCs) are the precursors of sperms and oocytes. Proper development of PGCs is crucial for the survival of the species. In many organisms, factors responsible for PGC development are synthesized during early oogenesis and assembled into the germ plasm. During early embryonic development, germ plasm is inherited by a few cells, leading to the formation of PGCs. While germline development has been extensively studied, how components of the germ plasm regulate PGC development is not fully understood. Here, we report that Dzip1 is dynamically expressed in vertebrate germline and is a novel component of the germ plasm in Xenopus and zebrafish. Knockdown of Dzip1 impairs PGC development in Xenopus embryos. At the molecular level, Dzip1 physically interacts with Dazl, an evolutionarily conserved RNA-binding protein that plays a multifaced role during germline development. We further showed that the sequence between amino acid residues 282 and 550 of Dzip1 is responsible for binding to Dazl. Disruption of the binding between Dzip1 and Dazl leads to defective PGC development. Taken together, our results presented here demonstrate that Dzip1 is dynamically expressed in the vertebrate germline and plays a novel function during Xenopus PGC development.

developmental biology↗
bioRxiv2023

Phase transition of maternal RNAs during vertebrate oocyte-to-embryo transition

The oocyte-to-embryo transition (OET) is regulated by maternal products stored in the oocyte cytoplasm, independent of transcription. How maternal products are precisely remodeled to dictate the OET remains an open question. In this work, we discover the dynamic phase transition of maternal RNAs during Xenopus OET. We have identified 863 maternal transcripts that transition from a soluble state to a detergent-insoluble one after oocyte maturation. These RNAs are enriched in the animal hemisphere and many of them encode key cell cycle regulators. In contrast, 165 transcripts, including nearly all Xenopus germline RNAs and some vegetally localized somatic RNAs, undergo an insoluble-to-soluble phase transition. This phenomenon is conserved in zebrafish. Our results demonstrate that the phase transition of germline RNAs influences their susceptibility to RNA degradation machinery and is mediated by the remodeling of germ plasm. This work thus uncovers novel remodeling mechanisms that act on RNAs to regulate vertebrate OET.

developmental biology↗
bioRxiv2022

Destin2: integrative and cross-modality analysis of single-cell chromatin accessibility data

We propose Destin2, a novel statistical and computational method for cross-modality dimension reduction, clustering, and trajectory reconstruction for single-cell ATAC-seq data. The framework integrates cellular-level epigenomic profiles from peak accessibility, motif deviation score, and pseudo-gene activity and learns a shared manifold using the multimodal input, followed by clustering and/or trajectory inference. We apply Destin2 to real scATAC-seq datasets with both discretized cell types and transient cell states and carry out benchmarking studies against existing methods based on unimodal analyses. Using cell-type labels transferred with high confidence from unmatched single-cell RNA sequencing data, we adopt four performance assessment metrics and demonstrate how Destin2 corroborates and improves upon existing methods. Using single-cell RNA and ATAC multiomic data, we further exemplify how Destins cross-modality integrative analyses preserve true cell-cell similarities using the matched cell pairs as ground truths. Destin2 is compiled as a freely available R package available at https://github.com/yuchaojiang/Destin2.

bioinformatics↗
bioRxiv2021

PTBP1/HNRNP I controls intestinal epithelial cell regeneration by maintaining stem cell survival and stemness

Properly controlled intestinal epithelial cell regeneration is not only vital for protection against insults from environmental hazards but also crucial for preventing intestinal cancer. Intestinal stem cells located in the crypt region provide the driving force for epithelial regeneration, and thus their survival and death must be precisely regulated. We show here that polypyrimidine tract binding protein 1 (PTBP1, also called heterogeneous nuclear ribonucleoprotein I, or HNRNP I), an RNA-binding protein that post-transcriptionally regulates gene expression, is critical for intestinal stem cell survival and stemness. Mechanistically, we show that PTBP1 inhibits the expression of PHLDA3, an AKT repressor, and thereby maintains AKT activity in the intestinal stem cell compartment to promote stem cell survival and proliferation. Furthermore, we show that PTBP1 inhibits the expression of PTBP2, a paralog of PTBP1 that is known to induce neuron differentiation, through repressing inclusion of alternative exon 10 to Ptbp2 transcript. Loss of PTBP1 results in a significant upregulation of PTBP2, which is accompanied by splicing changes in genes that are important for neuronal cell development. This finding suggests that PTBP1 prevents aberrant differentiation of intestinal stem cells into neuronal cells through inhibiting PTBP2. Our results thus reveal a novel mechanism whereby PTBP1 maintains intestinal stem cell survival and stemness through the control of gene function post-transcriptionally.

molecular biology↗
bioRxiv2020

Genome-wide circadian rhythm detection methods: systematic evaluations and practical guidelines

Circadian rhythms are oscillations of behavior, physiology, and metabolism in many organisms. Recent advancements in omics technology make it possible for genome-wide profiling of circadian rhythms. Here, we conducted a comprehensive analysis of seven existing algorithms commonly used for circadian rhythm detection. Using gold-standard circadian and non-circadian genes, we systematically evaluated the accuracy and reproducibility of the algorithms on empirical datasets generated from various omics platforms under different experimental designs. We also carried out extensive simulation studies to test each algorithms robustness to key variables, including sampling patterns, replicates, waveforms, signal-to-noise ratios, uneven samplings, and missing values. Furthermore, we examined the distributions of the nominal p-values under the null and raised issues with multiple testing corrections using traditional approaches. With our assessment, we provide method selection guidelines for circadian rhythm detection, which are applicable to different types of high-throughput omics data. Key pointsO_LIVarious methods have been developed for circadian rhythm detection on a genome-wide scale using omics technologies, yet there has not been a comprehensive summary and evaluation of all existing methods to date. C_LIO_LIUsing gold-standard circadian and non-circadian genes, we systematically evaluated the accuracy and reproducibility of seven existing algorithms for circadian rhythm detection on empirical datasets generated from various omics platforms. C_LIO_LIWe carried out extensive simulation studies to test each algorithms robustness to key variables, including sampling patterns, replicates, waveforms, signal-to-noise ratios, uneven samplings, and missing values. C_LIO_LIWe examined the distributions of the nominal p-values under the null and raised issues with multiple testing corrections using the Benjamini-Hochberg procedure due to gene-gene correlation and testing being overly conservative. C_LIO_LIWe provide method selection guidelines for circadian rhythm detection, which are applicable to different types of high-throughput omics data. C_LI

bioinformatics↗
bioRxiv2019

Minutes-timescale 3D isotropic imaging of entire organs at subcellular resolution by content-aware compressed-sensing light-sheet microscopy

Instant 3D imaging of entire organs and organisms at cellular resolution is a recurring challenge in life science. Here we report on a computational light-sheet microscopy able to achieve minute-timescale mapping of entire macro-scale organs at high spatial resolution, thereby overcoming the throughput limit of current 3D microscopy implementations. Through combining a dual-side confocally-scanned Bessel light-sheet illumination which provides thinner-and-wider optical sectioning of deep tissues, with a content-aware compressed sensing (CACS) computation pipeline which further improves the contrast and resolution based on a single acquisition, our method yields 3D images with high, isotropic spatial resolution and rapid acquisition improved by two-orders of magnitude. We demonstrate the imaging of whole brain ([~]400 mm3), entire gastrocnemius and tibialis muscles ([~]200 mm3) of mouse at subcellular resolution (0.5-m isovoxel) and ultra-high throughput of 5[~]10 minutes per sample. Various system-level cellular analyses, such as mapping cell populations at different brain sub-regions, tracing long-distance projection neurons over the entire brain, and calculating neuromuscular junction occupancy across whole muscle, were also readily enabled by our method.

bioengineering↗
bioRxiv2017

Adaptation in plant genomes: bigger isn’t better, but it’s probably different

Here we have proposed the functional space hypothesis, positing that mutational target size scales with genome size, impacting the number, source, and genomic location of beneficial mutations that contribute to adaptation. Though motivated by preliminary evidence, mostly from Arabidopsis and maize, more data are needed before any rigorous assessment of the hypothesis can be made. If correct, the functional space hypothesis suggests that we should expect plants with large genomes to exhibit more functional mutations outside of genes, more regulatory variation, and likely less signal of strong selective sweeps reducing diversity. These differences have implications for how we study the evolution and development of plant genomes, from where we should look for signals of adaptation to what patterns we expect adaptation to leave in genetic diversity or gene expression data. While flowering plant genomes vary across more than three orders of magnitude in size, most studies of both functional and evolutionary genomics have focused on species at the extreme small edge of this scale. Our hypothesis predicts that methods and results from these small genomes may not replicate well as we begin to explore large plant genomes. Finally, while we have focused here on evidence from plant genomes, we see no a priori reason why similar arguments might not hold in other taxa as well.

evolutionary biology↗
bioRxiv2017

Evolutionarily Conserved Alternative Splicing Across Monocots

One difficulty when identifying and analyzing alternative splicing (AS) events in plants is distinguishing functional AS from splicing noise. One way to add confidence to the validity of a splice isoform is to observe that it is conserved across evolutionarily related species. We use a high throughput method to identify junction based conserved AS events from RNA-Seq data across nine plant species including: five grass monocots (maize, sorghum, rice, Brachpodium and foxtail millet), plus two non-grass monocots (bananan and African oil palm), the eudicot Arabidopsis and the basal angiosperm Amborella. In total, 9,804 conserved AS events within 19,235 genes were identified conserved between 2 or more species studied. In grasses containing large regions of conserved synteny, the frequency of conserved AS events is twice that observed for genes outside of conserved synteny blocks. In plant-specific RS and RS2Z subfamilies, we observe both conservation and divergence of AS events after the whole genome duplication in maize. In addition, plant-specific RS and RS2Z subfamilies are highly connected with R2R3-MYB in splicing networks. Furthermore, we discovered that the network based on genes harboring conserved AS events is enriched for phosphatases, kinases and ubiquitylation genes, which suggests that AS may participate in regulating signaling pathways. These data lay the foundation for identifying and studying conserved AS events in the monocots, particularly across grass species, and this conserved AS resource identifies an additional layer between genotype to phenotype that may impact future crop improvement efforts.

bioinformatics↗